The University of Massachusetts Amherst
University of Massachusetts Amherst

Search Google Appliance


Optimizing the Design of 3D-printed Parts

Assistant Professor Wen Chen

Assistant Professor Wen Chen of the Mechanical and Industrial Engineering Department was the co-lead author of a trailblazing paper that describes innovative new research to use computational approaches for optimizing the design of 3D-printed parts. The paper was published in Science Advances and was written in collaboration with Lawrence Livermore National Lab (LLNL). Chen had led the experimental and mechanical testing work while a postdoc at LLNL and now was the first leading author affiliated with UMass Amherst. See articles in Technology.Org,, and Nanowerk.

The paper describes how LLNL researchers have designed a new class of 3D-printed lattice structures that combine lightweight and high stiffness, despite breaking a rule previously thought to be required to exhibit such properties. One of the new structures additionally displays perfectly uniform response to forces in all directions.

As Chen explained, “This research is about design and manufacturing of isotropic and stiff 3D architected materials. Our findings highlight the enhanced potential of topology optimization over traditional methods to design materials with unprecedented properties.”

The collaborators observed that stiff architected materials have been a long-sought goal in ultralight materials and architecture design. Materials with a stochastic microstructure, like foams, typically exhibit low mechanical stiffness, whereas lattices with a designed microarchitecture often show significantly improved stiffness.

The researchers noted that these architected materials have previously been designed by rule, using the so-called “Maxwell criterion” to ensure that their deformation is dominated by the stretching of their struts. Unfortunately, classical designs following this rule tend to be anisotropic, or not identical nor omnidirectional, with stiffness depending on the load orientation.

“Here,” the collaborators said, “the novelty of our research is that we have designed stiff and isotropic [identical and omnidirectional] lattices by a method called topology optimization, an approach based on continuum finite element analysis. Furthermore, we utilize the state-of-the-art additive manufacturing to fabricate out designed lattice structures and validate predictions of their performance and demonstrate that they are as efficient as those designed by rule, despite appearing to violate the Maxwell criterion.”

As described in the Science Advances paper, an LLNL team co-led by engineer Seth Watts used topology optimization software that Watts wrote to create two unique unit cell designs composed of micro-architected trusses, one of which was designed to have isotropic material properties. These new structures were then fabricated and tested, and were found to outperform the octet truss, a standard geometric pattern for 3D printed lattice structures that are stiff but anisotropic.

As explained in its article about the research, Chen tested these various samples at different densities to see what would happen when they were compressed at different angles to validate their isotropic properties. Chen said he was surprised by the results and that the research has "improved the promise" of replacing the classical octet truss design.

"It shows you can use this computational tool to design the structure to meet your target performance—this opens up a new design modus for architected materials," Chen said. "Secondly, it improves the mechanical efficiency of architected design. For environments where you may have complex stress states, you want to have it as isotropic as possible. This [research] expands the application of our lattices because in a real application you often need a material that can take loading from multiple directions."

The researchers said the work also proves that by using topology optimization, engineers can design new structures that outperform those created with traditional "design-by-rule" approaches. (October 2019)